Four tests were launched on the 184 unmodified sequences:

• One with the extended keyword list, and using Antikeyword list
• One with the extended keyword list, without using Antikeyword list
• One with the limited keyword list, and using Antikeyword list
• One with the limited keyword list, without using Antikeyword list

The results obtained show how much the keyword list can affect the software's performance.

These results show the efficiency of our keyword list. Indeed, the limited keyword list, which is the more natural keyword list fron the ccl list, misses some dangerous sequences and have more false postive.
We can see that the antikeyword list has not a great impact on the screening. It just corrects some false postive hits, which is nonetheless a good point as the weakness of our software is the number of false positives raised. On the other hand, this shows the importance of the construction of the keyword list. The number of sequences correctly flagged with the limited keyword list is only half the number of the sequences correctly flagged with the extended keyword list. Moreover, the extended keywordlist doesn't add much false postive compared to the limited one, so a keyword list as detailed as our keyword list (or more) should be used for a sequence screening software. The construction of this list is so a crucial point for the operation of the software.

Go back to tests and results page